This project proposes to develop a much easier and more accurate tool for crystallographers to fit RNA backbone into density, by combining methods and experience from mathematics, molecular graphics, crystallography, and structural bioinformatics. The details of RNA backbone conformation are critical to many of the biomedically important new roles being found for both large and small RNA molecules: specific aptamer binding, control of splicing, specificity of protein interactions in systems from SiRNA to ribosomes, and especially to a mechanistic understanding of ribozyme catalysis. However, the correct fitting of RNA backbone atoms into electron density maps at the resolutions typical for RNA or RNP crystal structures is very difficult: even with a simplified sugar pucker description, there are 6 variable dihedral angles per residue, and only the phosphate and the base can be seen really clearly. If the hydrogen atoms are added, then a substantial percentage of oligonucleotide residues in currently deposited structures show physically impossible steric clashes, indicating that refinement started from the wrong combination of angles. The multi-dimensional search problem for RNA backbone will be addressed with inverse kinematics and related methods used by the Snoeyink group to improve the search for protein backbone alternatives in protein design, modified to allow for the unusual nature of the constraints provided by the fairly precise but partial knowledge of phosphate and base positions and orientations. The necessary step of screening the possible geometrical solutions for molecular reasonableness will be provided by the Richardson group's all-atom contact analysis and quality-filtered database statistics, previously shown successful on the assessment and improvement of protein crystal structures and on RNA structural bioinformatics. Practical tools will be built onto the existing KiNG and/or Mage systems that already have capabilities for model and map display and for model manipulation and analysis. Usability will benefit from Richardson lab crystallographic and model correction experience and from beta-testing by interested RNA crystallographers;speed and robustness will benefit from the Snoeyink group's expertise with algorithms and good programming practice.